Estimating fatigue life of steel bridges using continuous response monitoring: Methodology and applications

Abstract

The fatigue life of structural steel bridges is governed by the time-history of stresses generated in-situ in its fatigue-critical structural details under service conditions. However, these stresses are often not directly and accurately measurable due to the complex geometry of structural elements or access restrictions. Therefore, there is a need for an approach to infer stresses at a detail using measurements taken away from the detail. Another practical issue is that instrumenting all structural details is infeasible owing high cost. An approach to infer stresses across the bridge only a limited number of sensors is therefore essential. This thesis aims to address the aforementioned two critical issues in monitoring-based fatigue life evaluation. It accomplishes this by investigating the following hypothesis: detailed finite element models of fatigue-critical connections and in-service strain measurements that capture the shear, flexure, and axial demands of the modelled connections can be combined to estimate accurately the in-situ hot spot/nominal stresses. This will enable much more reliable assessment of fatigue life than is possible by current methods. Proving this hypothesis will also permit expanding the approach to predicting hot spot/nominal stresses at uninstrumented connections by combining numerical models with real-time measurements from a few instrumented connections. This thesis focuses specifically on investigating this hypothesis on the fatigue-sensitive web-gap welded details in ladder-type bridge decks although the presented ideas are applicable to riveted/bolted connections in this type of bridges. The proposed approaches are evaluated using measurements from three full-scale bridges. Results illustrate that the methodology accurately predicts fatigue stress response. Results also confirm that the methodology can be utilised to infer stress time-histories at uninstrumented connections and to plan for retrofits. The study demonstrates that the proposed methodology is applicable for interpreting measurements from full-scale bridges, and can be integrated within a measurement interpretation platform for continuous bridge monitoring.